Notes on harmonic-Ricci flow on surface
Xiang-Zhi Cao · 2026
Read Notes on harmonic-Ricci flow on surface by Xiang-Zhi Cao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-05599.txt.
Introduction
Xiang-Zhi Cao's short note works on harmonic-Ricci flow over a surface that has a boundary. Harmonic-Ricci flow couples the stretching of a surface's geometry to a map into another space, and functionals are energy-like quantities tracked as the flow runs. Cao sets out to establish several formulas for those functionals in this boundary setting. The note is brief and purely mathematical, presenting the identities as groundwork for studying how such flows behave on surfaces with edges.
Catalog record
| CREATOR | Xiang-Zhi Cao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05599
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Notes on harmonic-Ricci flow on surface Xiang-Zhi Cao Abstract In this note, we want to establish several formulas about functionals along harmonic Ricci flow on surface with boundary. Keyword: harmonic ricci flow, surface with boundary. MSC: 58 1 Introduction and Preliminary As we know, Perelman [ 1 , 2 , 3 ] solved Poincare’s conjecture via Ricci flow. Later, Müller [ 4 ] combined Ricci flow with harmonic map heat flow, introducing harmonic-Ricci flow. Definition 1 (cf. [ 4 ] ) . The triples (M,g(t),\phi) is called harmonic-Ricci flow if \begin{cases}\frac{\partial g}{\partial t}=-2\mathrm{Ric}+2\alpha\nabla\phi\otimes\nabla\phi\\ \frac{\partial\phi}{\partial t}=\tau(\phi).\end{cases} (RH) In the past decades, harmonic-Ricci flow was an active research topic. One can refer to these works ( [ 5 , 6 , 7 , 8 , 9 ] ) for progress on harmonic-Ricci flow. There has been many results about the monotinicity quantity on closed Riemannian manifold such as Perelman’s F -functional and W -functional. One can also refer to these works ( [ 4 , 10 , 7 , 9 ] ). In the past decades, there are some progress about Ricci flow on manifold with boundary, such as ( [ 11 , 12 , 13 , 14 , 15 , 16 ] [ 17 ] ). In this paper, we consider pseudo-harmonic-Ricci flow, Definition 2 . The triples (M,g(t),\phi) is called psudo-harmonic-Ricci flow if \begin{split}\begin{cases}\frac{\partial g}{\partial t}=-2\mathrm{Ric}+2\alpha\nabla\phi\otimes\nabla\phi\\ \tau(\phi)=0\\ \frac{\partial\phi}{\partial t}=0.\end{cases}\end{split} (1.1) Obviously, psudo-harmonic-Ricci flow is special harmonic Ricci flow. Our main purpose of this paper is to obtain monotonicity formula of several functional, such as entropy functional (see ( 2.1 )), Perelman type \mathcal{F} -entropy, Perelman type \mathcal{W} -entropy on surface with boundary along the flow ( 1.1 ). Notations: We define \tau:=T-t . We write \tau(\phi) for the tension field of the map \phi. By abuse of notations, these two kinds of \tau can be clarified from the context. As in literature, we also use the notations \operatorname{Sc}:=\mathrm{Ric}-\alpha\nabla\phi\otimes\nabla\phi , S=\mathrm{tr\,}_{g}(\operatorname{Sc}) , we use the symbol R to denote the scalar curvatture of the metric g . we use \boldsymbol{n} to denote the unit normal vector field at the point of the boundary. We use \mathrm{d}v to denote the volume measure, and \mathrm{d}A to represent the measure of the boundary. As we know, for surface the formula \mathrm{Ric}=-2Rg holds. We denote \bar{S} by \frac{\int_{M}S\mathrm{d}v}{\int_{M}\mathrm{d}v} . By the results of harmonic-Ricci flow, using the above notations, we recall a result Lemma 1 (cf. [ 4 ] ) . Let (M,g(t),\phi) be psudo-harmonic-Ricci flow (cf. ( 1.1 )) . Then \begin{split}\frac{\partial S}{\partial t}=\Delta_{g}S+2\left\lVert Sc\right\rVert^{2}+2\alpha\left\lVert\tau_{g}(\phi)\right\rVert^{2}.\end{split} Our main results are Theorem 1 , Theorem 4 and Theorem 7 . Throughout this paper, we assume \alpha to be positive constant. 2 Entropy on surface with boundary In this section, we generalize some results established in [ 12 ] to pseudo-harmonic-Ricci flow. Specifically, we introduce the modified entropy \begin{split}E_{\partial}(t)=\int_{M}S\log S\mathrm{d}v-\log(\bar{S})\int_{M}S\mathrm{d}v,\end{split} (2.1) provided that S>0. Theorem 1 . Let (M,g(t),\phi),t\in[0,T) be a pseudo-harmonic-Ricci flow with S>0 on compact surface with boundary \partial M , R(g_{0})>0. We further assume the Neumann type boundary condition \frac{\partial R}{\partial\boldsymbol{n}}=0,\frac{\partial\phi}{\partial\boldsymbol{n}}=0 . Let f be a smooth function on M solving \displaystyle\begin{cases}S+\Delta f=\bar{S}\\ \frac{\partial f}{\partial\boldsymbol{n}}=0,\end{cases} then \displaystyle\frac{\mathrm{d}}{\mathrm{d}t}E_{\partial}(t) \displaystyle=-\int_{M}\alpha\left\lVert\nabla\phi(\nabla f)\right\rVert^{2}\mathrm{d}v \displaystyle-\int_{M}S\left\lVert\nabla S-\nabla\log S\right\rVert^{2}-2\int_{M}\left\lVert\nabla^{2}f-\frac{\Delta f}{2}g\right\rVert^{2}\mathrm{d}v \displaystyle-\int_{\partial M}\Pi(\nabla f|_{\partial M},\nabla f|_{\partial M})\mathrm{d}A. In particular, if the boundary is geodesic convex, then the entropy E_{\partial}(t) is non-decreasing. Remark 1 . Should the conditon of pseudo-harmonic-Ricci flow be replaced by harmonic Ricci flow, the Theorem may no longer hold. Proof. Adjusting the proof of [ 12 ] to our setting, we have \begin{split}\frac{\mathrm{d}}{\mathrm{d}t}E_{\partial}(t)&=\int_{M}\partial_{t}S(1+\log S)-S^{2}\log S\,\mathrm{d}v-\partial_{t}(\log\bar{S}\int_{M}S\,\mathrm{d}v)\\ &=\int_{M}(\Delta S+2\left\lVert Sc\right\rVert^{2}+2\alpha\left\lVert\tau(\phi)\right\rVert^{2})(1+\log S)-S^{2}\log S\mathrm{d}v-v(M)\bar{S}^{2}\,\\ &=\int_{M}(\Delta S\log S+2\left\lVert S\right\rVert^{2})\mathrm{d}v+\int_{M}2\alpha\left\lVert\tau(\phi)\right\rVert^{2}(1+\log S)\mathrm{d}v-v(M)\bar{S}^{2}\,\\ &=\int_{M}(-S\left\lVert\nabla\log S\right\rVert^{2}+2\left\lVert S\right\rVert^{2})\mathrm{d}v+\int_{M}2\alpha\left\lVert\tau(\phi)\right\rVert^{2}(1+\log S)\mathrm{d}v-v(M)\bar{S}^{2}.\end{split} Observe that \displaystyle\int_{M}\left\lVert S\right\rVert^{2}\mathrm{d}v=\int_{M}(\Delta f)^{2}\mathrm{d}v+v(M)\bar{S}^{2}. Using this, we deduce that \displaystyle\frac{\mathrm{d}}{\mathrm{d}t}E_{\partial}(t)=-\int_{M}(S\left\lVert\nabla\log S\right\rVert^{2}-2\left\lVert\Delta f\right\rVert^{2})\mathrm{d}v+\int_{M}2\alpha\left\lVert\tau(\phi)\right\rVert^{2}(1+\log S)\mathrm{d}v. By integration by parts and using f_{\nu}=0 , we get \begin{split}&\quad\,\,\int_{M}S\|\nabla f\|^{2}-2(\Delta f)^{2}+S\|\nabla\log S\|^{2}\,\mathrm{d}v\\ &=\int_{M}S\|\nabla f\|^{2}+2g(\nabla f,\nabla\Delta f)+S\|\nabla\log S\|^{2}\,\mathrm{d}v\\ &=\int_{M}\,S\|\nabla f-\nabla\log S\|^{2}\,\mathrm{d}v.\end{split} (2.2) Furthermore, using Reilly formula gives \int_{M}\,2\left(\Delta f\right)^{2}-R\|\nabla f\|^{2}-2\left\|\nabla^{2}f\right\|^{2}\,\mathrm{d}v=2\int_{\partial M}\,\left(\nabla_{\partial M}(f|_{\partial M}),\nabla_{\partial M}(f|_{\partial M})\right)\,\mathrm{d}A. (2.3) Adding ( 2.2 ) and ( 2.3 ), we conclude that \displaystyle\int_{M}(S\left\lVert\nabla\log S\right\rVert^{2}-2\left\lVert\nabla^{2}f\right\rVert^{2}-\alpha\left\lVert\nabla\phi\right\rVert^{2}\left\lVert\nabla f\right\rVert^{2}\mathrm{d}v \displaystyle= \displaystyle-\int_{M}S\left\lVert\nabla S-\nabla\log S\right\rVert^{2}\mathrm{d}v-\int_{\partial M}\Pi(\nabla f|_{\partial M},\nabla f|_{\partial M})\mathrm{d}A. Combining these yields \displaystyle\frac{\mathrm{d}}{\mathrm{d}t}E_{\partial}(t)= \displaystyle\int_{M}2\left\lVert\Delta f\right\rVert^{2}\mathrm{d}v+\int_{M}2\alpha\left\lVert\tau(\phi)\right\rVert^{2}(1+\log S)\mathrm{d}v \displaystyle\int_{M}-2\left\lVert\nabla^{2}f\right\rVert^{2}-\alpha\left\lVert\nabla\phi\right\rVert^{2}\left\lVert\nabla f\right\rVert^{2}\mathrm{d}v \displaystyle-\int_{M}S\left\lVert\nabla S-\nabla\log S\right\rVert^{2}\,\mathrm{d}v-\int_{\partial M}\Pi(\nabla f|_{\partial M},\nabla f|_{\partial M})\,\mathrm{d}A. It is now clearly that the claim of the Theorem follows immediately. \hfill\Box 3 Perelman type \mathcal{F} -entropy Perelman’s F -entropy and W -entropy are defined by respectivly \displaystyle F(g,f)=\int_{M}(R+\left\lVert\nabla f\right\rVert^{2})e^{-f}\mathrm{d}v, and W(g,f)=\int_{M}\tau(R+\left\lVert\nabla f\right\rVert^{2})+f-m)(4\pi\tau)^{-1}e^{-f}\mathrm{d}v. Later,m \ddot{u} ller generalize Perelman’s F -entropy and W -entropy to the case of harmonic-Ricci flow on compact closed manifold. They introduced the following functional \begin{split}\mathcal{F}(g,\phi,f)=\int_{M}(R-\alpha\left\lVert\nabla\phi\right\rVert^{2}+\left\lVert\nabla f\right\rVert^{2})e^{-f}\mathrm{d}v,\end{split} (3.1) and \mathcal{W}(g,\phi,f)=\int_{M}\tau(R-\alpha\left\lVert\nabla\phi\right\rVert^{2}+\left\lVert\nabla f\right\rVert^{2})+f-m)(4\pi\tau)^{-1}e^{-f}\mathrm{d}v, on m -dimensional manifold (M^{m},g). Moreover, they established monotonicity formulas of the functionals just mentioned. In this paper we try to extend these results to the case of surface with boundary. In the case of Ricci flow on surface with boundary, there is already some progress on surface with boundary, such as Lemma 2 (cf. [ 11 ] ) . Let (M^{2},g) be surface with boundary. Denote \delta g_{ij}=v_{ij},\delta f=h,g^{ij}v_{ij}=v . Then \begin{split}&\delta\int_{M}(R+\left\lVert\nabla f\right\rVert^{2})e^{-f}\mathrm{d}v\\ =&\int_{M}e^{-f}\left[-v^{ij}\left(R_{ij}+\nabla_{i}\nabla_{j}f\right)+\right(\frac{v}{2}-h\left)\left(2\Delta_{g}f-\left|\nabla f\right|^{2}+R_{g}\right)\right]\,\mathrm{d}v\\ &-\int_{\partial M}\left[\frac{\partial v}{\partial\boldsymbol{n}}+\left(v-2h\right)\frac{\partial f}{\partial\boldsymbol{n}}\right]e^{-f}\,\mathrm{d}v\\ &+\int_{\partial M}e^{-f}\nabla_{i}v_{ij}\eta^{j}\,\mathrm{d}A-\int_{\partial M}\nabla_{j}e^{-f}v_{ij}\eta^{i}\,\mathrm{d}A.\end{split} (3.2) By A routime compuatation , we can get Lemma 3 . Let (M,g) be compact manifold with boundary. Let \delta g_{ij}=v_{ij},\delta f=h,g^{ij}v_{ij}=v,\delta\phi=\theta , then \displaystyle\delta\int_{M}(-\alpha\left\lVert\nabla\phi\right\rVert^{2})e^{-f}\mathrm{d}v \displaystyle= \displaystyle\int_{M}2\alpha\theta^{\lambda}\left(\Delta\phi^{\lambda}-\left\langle\nabla\phi^{\lambda},\nabla f\right\rangle\right)e^{-f}\mathrm{d}v+\int_{M}-\nu^{ij}\bigg(\alpha\nabla_{i}\phi^{\lambda}\nabla_{j}\phi^{\lambda}\bigg)e^{-f}\mathrm{d}v \displaystyle+\alpha\left\lVert\nabla\phi\right\rVert^{2}(\frac{1}{2}\mathrm{tr\,}_{g}\nu-h)e^{-f}\mathrm{d}v \displaystyle+\int_{\partial M}2\alpha\left\langle\nabla_{\boldsymbol{n}}\phi,\theta\right\rangle e^{-f}\mathrm{d}A. Proof. A routine computation shows that \begin{split}&\delta\int_{M}\left\lVert\nabla\phi\right\rVert^{2}e^{-f}\mathrm{d}v\\ =&\int_{M}2g^{ij}\nabla_{i}\phi^{\lambda}\nabla_{j}\theta^{\lambda}e^{-f}\mathrm{d}v+\int_{M}\bigg(-\nu^{ij}\nabla_{i}\phi^{\lambda}\nabla_{j}\phi^{\lambda}\bigg)e^{-f}\mathrm{d}v+\left\lVert\nabla\phi\right\rVert^{2}(\frac{1}{2}\mathrm{tr\,}_{g}\nu-h)e^{-f}\mathrm{d}v.\\ \end{split} (3.3) Integration by parts , the proof of the Lemma is complete. \hfill\Box Lemma 2 and Lemma 3 imply that Theorem 2 . Let (M^{2},g) be surface with boundary, using the above notations, then \begin{split}&\delta\mathcal{F}(g,\phi,f)\\ =&\delta\int_{M}(R-\alpha\left\lVert\nabla\phi\right\rVert^{2}+\left\lVert\nabla f\right\rVert^{2})e^{-f}\mathrm{d}v\\ =&2\alpha\int_{M}\theta^{\lambda}\left(\Delta\phi^{\lambda}-\left\langle\nabla\phi^{\lambda},\nabla f\right\rangle\right)e^{-f}\mathrm{d}v\\ &+\int_{M}-\nu^{ij}\bigg(\mathrm{Ric}_{ij}+\nabla_{i}\nabla_{j}f-\alpha\nabla_{i}\phi^{\lambda}\nabla_{j}\phi^{\lambda}\bigg)e^{-f}-\alpha\left\lVert\nabla\phi\right\rVert^{2}(\frac{1}{2}\mathrm{tr\,}_{g}v-h)e^{-f}\mathrm{d}v\\ &+\int_{\partial M}2\left\langle\nabla_{\boldsymbol{n}}\phi,\theta\right\rangle e^{-f}+\left(\frac{\mathrm{tr\,}_{g}v}{2}-h\right)\left(2\Delta_{g}f-\left|\nabla f\right|^{2}+R_{g}\right)\mathrm{d}A\\ &-\int_{\partial M}\left[\frac{\partial v}{\partial\boldsymbol{n}}+\left(\mathrm{tr\,}_{g}v-2h\right)\frac{\partial f}{\partial\boldsymbol{n}}\right]e^{-f}\,\mathrm{d}A\\ &+\int_{\partial M}e^{-f}\nabla_{i}v_{ij}\eta^{j}\,\mathrm{d}A-\int_{\partial M}\nabla_{j}e^{-f}v_{ij}\eta^{i}\,\mathrm{d}A.\end{split} (3.4) We now proceed to simplify the formula given by ( 3.4 ). Let us consider the evolution equations given by \left\{\begin{array}[]{l}\frac{\partial}{\partial t}g_{ij}=-2\left(R_{ij}+\nabla_{i}\nabla_{j}f-\alpha\nabla_{i}\phi^{\lambda}\nabla_{j}\phi^{\lambda}\right)\quad\mbox{in}\quad M\times\left(0,T\right)\\ k_{g}\left(\cdot,t\right)=\psi\left(\cdot\right)\quad\mbox{on}\quad\partial M\times\left(0,T\right)\\ \frac{\partial f}{\partial t}=-\Delta_{g}f+\left|\nabla f\right|^{2}-R_{g}+\alpha\left\lVert\nabla\phi\right\rVert^{2}\quad\mbox{in}\quad M\times\left(0,T\right)\\ \frac{\partial}{\partial\boldsymbol{n}}f=0\quad\mbox{on}\quad\partial M\times\left(0,T\right)\\ \frac{\partial\phi}{\partial t}=\Delta\phi^{\lambda}-\left\langle\nabla\phi^{\lambda},\nabla f\right\rangle,\end{array}\right. (3.5) on surface M with boundary \partial M . By the diffeomorphism generated by the vector field \nabla f , we find the systems ( 3.5 ) is equivalent to the following systems: \left\{\begin{array}[]{l}\frac{\partial}{\partial t}g_{ij}=-2\left(R_{ij}-\alpha\nabla_{i}\phi^{\lambda}\nabla_{j}\phi^{\lambda}\right)\quad\mbox{in}\quad M\times\left(0,T\right)\\ k_{g}\left(\cdot,t\right)=\psi\left(\cdot\right)\quad\mbox{on}\quad\partial M\times\left(0,T\right)\\ \frac{\partial f}{\partial t}=-\Delta_{g}f-R_{g}+\alpha\left\lVert\nabla\phi\right\rVert^{2}\quad\mbox{in}\quad M\times\left(0,T\right)\\ \frac{\partial f}{\partial\boldsymbol{n}}=0\quad\mbox{on}\quad\partial M\times\left(0,T\right)\\ \frac{\partial\phi}{\partial t}=\tau(\phi).\end{array}\right. (3.6) By Lemma 2 , we can show that Theorem 3 . Under the flow ( 3.6 ), we have \begin{split}&\frac{d}{dt}\mathcal{F}(g,\phi,f)\\ =&2\int_{M}\left\|R_{ij}+\nabla_{i}\nabla_{j}f-\alpha\nabla_{i}\phi\otimes\nabla_{j}\phi\right\|^{2}e^{-f}\,\mathrm{d}v+2\alpha\int_{M}\left\lVert\tau(\phi)-\left\langle\nabla\phi,\nabla f\right\rangle\right\rVert^{2}e^{-f}\mathrm{d}v\\ &+\int_{\partial M}\left(k_{g}R_{g}-2k_{g}^{\prime}\right)e^{-f}\,\mathrm{d}A+2\int_{\partial M}k_{g}\left\|\nabla^{\top}f\right\|^{2}e^{-f}\,\mathrm{d}A,\\ &+\int_{\partial M}2\left\langle\nabla_{\boldsymbol{n}}\phi,\theta\right\rangle e^{-f}\mathrm{d}A,\end{split} where \nabla^{\top}f denotes the component of \nabla f tangent to \partial M , k_{g}^{\prime} denotes the derivatives of k_{g} with respect to the time t. Proof. We can carry over verbatim the proof of [ 11 ] to our set-up. It suffices to note that in our case the variations is given by v_{ij}=\delta g_{ij}=-2\left(R_{ij}-\alpha\left\lVert\nabla\phi\right\rVert^{2}+\nabla_{i}\nabla_{j}f\right),\quad h=\delta f=-\Delta_{g}f-R. The remained proof is omitted. \hfill\Box Obviously, Theorem 3 implies the following result Theorem 4 . The functional \mathcal{F}(g,\phi,f) given in ( 3.1 ) is non-decreasing along the flow ( 3.5 ) on the interval (0,T) . 4 Perelman type \mathcal{W} -entropy Next, we consider Perelman’s \mathcal{W} -functional on surface M with boundary \partial M , which is defined as \mathcal{W}_{\text{Perelman}}\left(g,f,\tau\right)=\frac{1}{4\pi\tau}\int_{M}\left[\tau\left(\left|\nabla f\right|^{2}+R_{g}\right)+f-2\right]e^{-f}\,dA_{g}. Recall that k_{g}^{\prime} denotes the derivatives of k_{g} with respect to the time t . Provided that k_{g}\geqslant 0 and k_{g}^{\prime}=\psi^{\prime}\leqslant 0 , we recall the following result in the literature Theorem 5 (cf. [ 11 ]